Wall Street Asks What AI Bought. Can You Answer? (Nov 6, 2026)

🎧 Listen ad-free:


🌐 Daily Briefing: https://djamgamind.com/daily
📚 Research Hub: https://djamgamind.com/pdfs

🔍 Keywords: AI Costs, AI ROI, AI Budgets, Nasdaq, AI Stocks, Apple, Deloitte, Anthropic IPO, METR, Gemini 4 Carbon, GPT-6 Cyber, AI Cyberattacks

Today’s question: The AI bill is coming due. Can you show what yours bought?

Summary:
The Nasdaq fell 5.5% on Thursday, its worst day since April 2025, as investors decided the AI buildout will be slower and more expensive. Apple cut some engineers’ daily AI budget from $300 to $200. Deloitte finds 81% of companies use AI to decide faster, but only 25% say it has transformed the business. Inside Anthropic, METR estimates AI has roughly doubled the pace of progress. And while OpenAI armed defenders with GPT-6 Cyber, hackers ran AI agents against more than 200 organizations in 11 days.

Important Topics:

  • The market asks for receipts: Nasdaq -5.5%, chips -12%, Asia follows.

  • The token budget: Apple’s AI rationing and Deloitte’s ROI gap.

  • Where AI does pay: METR inside Anthropic, and the IPO filing’s numbers.

  • The cost of defense: GPT-6 Cyber and AI-run hacking.

  • Today’s corners: Cert Corner, Red Flag or Green Light?, and the Boardroom Call.


📉 The market asks for receipts

Axios AI+ · Fox Business · The National · TBPN

  • The close: Nasdaq -5.52%, S&P 500 -4.36%, Dow -1,833 points (Fox Business). The Nasdaq 100 fell 6.4%, its steepest drop since March 2020, and the SOX chip index fell 12.3% (Axios AI+).

  • The suppliers took the hit: AMD -17.3%, Dell and Ciena -15%, Vertiv -14%, Coherent -13%, Quanta -12% (Axios AI+).

  • Asia on Friday: the Nikkei slid about 7% at the open, and South Korea’s Kospi fell more than 8%, triggering a 20-minute trading halt. Losses eased later (The National).

  • Why: “The market’s interpretation is that the AI infrastructure buildout will become slower and more expensive, which is rewriting growth expectations across the whole supply chain,” says eToro’s Jakub Rochlitz (Fox Business).

  • The question investors keep asking: “what comes after AI coding?” (TBPN). Axios is watching whether the volatility makes Anthropic and OpenAI rethink their IPOs, Betting odds on Kalshi of Anthropic filing by Nov 30 fell from 84% to 60% (Techpresso), and Canada’s Cohere says its IPO timing is uncertain (WSJ headline via TBPN).

Why it matters: A falling market doesn’t change what AI can do for you on Monday. It changes who funds it. Check the financial health of the AI vendors you depend on, and your exit terms.

🤝 PARTNER SLOT
This space is reserved for one enterprise AI partner each day: AI cost control, model routing and security for the leaders who read this briefing. Sponsorship options and media kit →

💰 The token budget

Techpresso · The Deep View · The Batch

  • Apple rations AI: in October Apple cut some engineers’ daily AI budget from $300 to $200 and is weighing $100. Some non-engineering teams could drop from $100 to as little as $25 (Techpresso, citing Bloomberg).

  • Deloitte’s gap: 81% of enterprises use AI to decide faster, but only 25% say it has meaningfully transformed the business. Revenue gains from AI rose to 28%, from 20% in January (The Deep View).

  • Deloitte’s Jim Rowan: “The real ROI comes not only from doing existing work faster, but from redesigning work altogether.”

  • What gets measured: about 40% of companies measure AI by hours saved; only 27% track net-new revenue (Wharton, via The Deep View).

  • Andrew Ng: the bottleneck “is not coding, but product management, workflow redesign, and other tasks such as judgement tasks” (The Batch).

The Number: 23%. Gemini 4 Carbon will cost the same per token as Claude Opus 5.5 once its discount ends, yet it costs 23% less per task because it uses 29% fewer output tokens (The Batch). Token efficiency is now part of the price. At the low end, Thinking Machines says its open-weight Inkling 2 Small costs about $1.18 per task (Techpresso, company claim).

Why it matters: When a CFO starts rationing tokens, a flat cap is the easy answer and usually the wrong one. Measure cost per task and per outcome, and route simple work to cheaper models.

📈 Where AI does pay: inside Anthropic

The Rundown AI · Axios AI+ · The Batch

  • METR’s estimate: after 11 weeks of inside access, the evaluation nonprofit METR estimates Anthropic’s progress runs about 2x its 2024 to 2025 pace (range 1.3x to 3x), with AI credited for around 80% of the speedup (The Rundown AI).

  • Redacted: Anthropic held back most of the strongest evidence. Co-founder Sam McCandlish said it was to avoid “handing out a roadmap for others to go faster.” (The Rundown AI)

  • The filing’s numbers: revenue of $16.3 billion in the first half of 2026, up from $1.2 billion; a $974 million operating profit in Q2; compute fell from 251% of revenue to 48% (The Batch, from the S-1).

  • AI writing the code: Claude wrote more than 90% of contributions to Anthropic’s codebase in September. The filing warns this “could also pose a considerable safety risk.” (The Batch)

  • The listing: ticker ANT on the Nasdaq, possibly as soon as the week before Thanksgiving, seeking roughly $100 billion (WSJ via The Batch).

Why it matters: This is the other side of Deloitte’s gap. The return shows up where a company rebuilt how the work gets done, not where it handed out licenses.

🛡️ The cost of defense

The Deep View · Axios AI+ · BleepingComputer

  • GPT-6 Cyber: OpenAI’s model for defenders scores 48.5% on ExploitGym, up from 32.1% for its predecessor. OpenAI says it found 81 previously unknown “high-severity” Firefox flaws and stayed clear of “out-of-scope services” in 99.93% of tests (The Deep View).

  • AI-run hacking: a China-linked group CrowdStrike calls Vapor Panda used an AI tool called Artex, with DeepSeek and a commercial model, to target more than 200 organizations in 14 countries in 11 days (Axios AI+). At least nine were compromised (BleepingComputer).

  • CrowdStrike’s Adam Meyers: “It was doing the stuff that you, as a pen tester, would do,” but “at a scale and a speed that I think a human would not be able to do.” (Axios AI+)

  • Agents fall for scams: a test that rebuilt real fraud sites found AI shopping and banking assistants can be fooled up to 35% of the time (research summaries via Techpresso).

Why it matters: Attackers now rent speed cheaply. If you buy AI security testing, buy it with a written scope, a list of systems it must never touch, and a kill switch.


🎯 Today’s corners

🎓 Cert Corner: Your customer-support assistant sends every request to the most expensive AI model, and the bill doubled in six months. What’s the best first step to cut cost without hurting quality?

A) Switch everything to the cheapest model

B) Measure cost per resolved request, then route simple requests to a smaller model

C) Turn off logging

D) Allow longer answers so users ask fewer follow-ups

🚩 Red Flag or Green Light? A vendor offers an “autonomous pen-testing agent” to probe your network overnight. The contract has no written scope, no list of systems it must not touch, and no way to stop it mid-run. 🚩 or ✅?

🏛️ The Boardroom Call (cost): Your AI tool bill doubled in six months, and Apple just cut some engineers’ daily AI budget from $300 to $200.

A) A flat daily cap for everyone

B) Keep spending, but each team shows one measured business result per quarter

C) Freeze new AI spend until the market settles

D) Move routine work to cheaper models

Answers below. Tell us yours in the comments.

Strategic C-Suite Takeaways

  1. Measure AI in outcomes, not hours saved or tokens used.

  2. Set a cost per task and route work by difficulty.

  3. Redesign one workflow end to end before buying more seats.

  4. Check your AI vendors’ financial health and your exit terms.

  5. Put scope, exclusions and a kill switch in any AI security-testing contract.

What Else Happened in AI on November 6, 2026?

  • OpenAI released Images 2.6, which turns prompts, photos or sketches into GIFs and makes images 50% faster. The Rundown AI

  • Google’s Gemini 4 Carbon took first place on Artificial Analysis’ Intelligence Index; access is limited to selected companies for now. The Batch

  • Perplexity launched Teammates, AI agents that each own an ongoing project. The Rundown AI

  • Anthropic launched Personal Finance for Claude in beta: read-only access to U.S. bank and card accounts through Plaid, no money movement. Techpresso

  • OpenAI pledged to pay the full power and infrastructure costs of its data centers under a “National Community Compact.” Techpresso

  • Google cut its Gemini Omni video price by 30% with 1.2 Flash. The Rundown AI


✅ Our take on today’s corners

🎓 Cert Corner: B. Measure first, then route by difficulty. The cheapest model everywhere risks quality, turning off logs removes the data you need, and longer answers usually cost more. Practice thousands more in DjamgaTech.

🚩 Red Flag. Tools like this are what Vapor Panda turned against 200 organizations. Before any AI probes your network: written scope, excluded systems, a kill switch, full logs and a scheduled test window. 📘 AI security controls are a full domain of the AWS AI Practitioner exam: our 2026 guide.

🏛️ B. Keep spending, but tie it to one measured result per team, and use cheaper models for routine work (D) to stretch the budget. A flat cap hits your heaviest users, who are often your most productive, and a freeze reacts to Wall Street instead of your own numbers. Your org does it differently? Tell us. 📘 Measuring AI’s business value is a core topic of Google’s Gen AI Leader exam: our 2026 guide.

🏛️ The Boardroom Call is open for one sponsor. AI cost and model-routing companies: djamgamind.com/partners


🤝 Advertise with AI Unraveled & DjamgaMind

Put your enterprise AI brand in front of the CTOs, CISOs and engineering leaders who make AI buying decisions, daily, in English and French, on podcast apps, YouTube and Substack. Start by sponsoring one daily corner for four weeks.
Sponsorship options & media kit → · Email: etienne_noumen@djamgamind.com

🔗 Tools from the show

🎧 Prefer it commercial-free? Subscribe on Apple Podcasts.

⚗️ PRODUCTION NOTE: We Practice What We Preach.
AI Unraveled is researched and produced by Etienne Noumen, P.Eng. (Calgary, Canada) using a hybrid “Human-in-the-Loop” workflow.



Read original on Substack

Related articles

Level up your Tech Career with DjamgaTech PRO

Download the DjamgaTech PRO App on the Apple App Store to access hundreds of interactive prep questions, flashcards, and quizzes cleanly without ads.

Djamgatech PRO App
Download DjamgaTech PRO on the App Store 🚀

← All articles